Nobel Physicist John Martinis on Raising Curious Kids, Learning with AI, and Why You Shouldn't Chase a Nobel

Genius Born or Made? — Himanshu Gupta in conversation with John Martinis (Nobel Laureate in Physics). Source: episode transcript provided by the host. Timestamps are at 1.25× playback speed to match the delivered video (1.25x timestamp = raw × 0.8). Speaker labels preserved. This is the faithfulness source of record: every quote on the episode page must appear verbatim below.

Himanshu Gupta (00:02.68) John, welcome to the show.

John Martinis (00:04.47) Yeah, thanks for inviting me.

Himanshu Gupta (00:06.58) Well, I still remember our conversations at within the Davos Congress Center and feel the insights that you share really intrigued me. And I thought our audience, is primarily parents of young kids and even youngsters can benefit from those insights as well. So I'm going to start with the first question that I start with my guest with and this is even more relevant for you being the Nobel Laureate. When it comes to your life, how much do you think your excellence was built versus acquired over a period of time?

John Martinis (00:41.82) Yeah, so that's really kind of a hard question to answer. It's because, you know, many, many things contributed to them. know, genetics is also important. was seen, I think I was born to have a mind to like physics. Let me explain that first. I actually have a very poor memory. So learning languages is particularly hard for me. When I was in France, I just had a hard time. But anyway, I have a bad memory. But this is in fact really good for physics because in physics, have a few basic concepts and formulas to memorize. And then you tend to derive everything. And the problem students have with physics is they try to memorize things. And the physics community really tries to teach people to think and understand the concepts and to derive things. And that's the way I've always approached physics, in part because I have a bad memory, bad with languages, for example. And thus, I naturally had to learn physics by understanding the concepts. And I fortunately had very good teachers that, like a high school physics teacher was great. I took honors physics at Berkeley as the first two years. They taught it really well from this point of view. And then that very much helped me. But going a little bit more to the environment, I think what was very good is my parents did something that was very clever to train me as a scientist. My mom was a homemaker. She graduated high school. My dad. He's a fireman. He basically got a GED for high school because of family situations. My dad's very technical. But they didn't come from science. But what they did is, you know how when you get around the family dinner table and, what did you do today? OK, and then the kids roll their eyes and they don't say very much. OK, but what my parents did is very clever. They asked.

John Martinis (02:37.63) did you ask any questions today? And then, okay, you what did you ask and what did you learn, et cetera, et But I was very much grew up in an environment where I was supposed to ask questions. And you know, that was deemed a good thing. And of course that's essential for being a scientist. You know, I go to a visit conference and people give talks. I always have lots of questions to ask. And I'm kind of not shy to ask it. think most physicists are shy. I'm not shy. You know, if I look bad because I don't understand something, yeah, so be it. But usually I ask a question because there's, you know, something a little bit subtle about it that needs to be understood and, you know, kind of start the conversation. And because I need to understand what's being presented in kind of my logical framework that I understand physics. But this is really good. Now what happens for the parents out there is kids are naturally curious, but they kind of learn over time that adults get annoyed if they ask too many questions. And you have to, you know, I understand that problem, but I think you have to, you know, make sure that you, you know, spend the time. So, you know, when our kids are growing up, the nice thing is that I could, you know, explain things at depth as far as the kids wanted to go for physical reality. And then if there was anything related to biology, that's my wife's expertise and she could do the same. So we had most things covered, maybe not business, but okay, most scientific things covered. And I think that's a good thing. But you know, I think for most of you, maybe you don't know science, but to emphasize that maybe, okay, let's do a internet search and find out what the answer to your question is. But yeah, you have to, you have to encourage this. I think that's really critical.

Himanshu Gupta (04:11.99) Okay, so a lot to unpack, John. You were born with poor memory. And as a result, you had no other choice but to lean towards physics. And then you had teachers, your parents did something unique, and that allowed you to become what you are. A lot to unpack there. Number one, when did you realize as a kid that you had poor memory? And how did that feel?

John Martinis (04:33.60) I don't know if I ever put it that way when I was a child. what happened is, I generally did well in all my classes, especially math and science. But when I took my language classes, say in high school, there was a real difficulty. And then later on when I lived in France for a few years, I saw that. And I always knew my memory wasn't the greatest. But that's the way you are and you live with it. But I'm gonna say I didn't do well in all of my subjects. I got A's in high school, but some subjects I had to work hard and it was just more difficult. And again, thinking back along it, a lot of that was memorable.

Himanshu Gupta (05:15.02) I see. And then, so in high school you realized, yeah.

John Martinis (05:18.53) I also want to think of things that thinking logically about things. My mom and dad were this way, especially my dad. He's very mechanically oriented. And when we talk about building things in the garage, he would always kind of explain things on a principles basis, more so than, you you just do it this way, which again, that trained me to be a physicist, even though we didn't have that education.

Himanshu Gupta (05:40.17) So that's an interesting point. Your dad explaining concepts to you from a principles basis. Give us an example of how would you explain a concept to your kid.

John Martinis (05:48.70) Well, just going back to my dad, he did electrical work, okay? And on the car, there was a hot wire and there was ground, okay? And he could explain a lot of that. You know electricity is more complicated than that, but for the kinds of things he was building, that was fine. And that's the way he explained it to me and he could get things done with. So it's an example, you know, he taught me the principle that was, you know, kind of on the way to being right. So, you know, in terms of thinking about principles, as I think about building a quantum computer, which is I'm doing right now, I tend to do a deep dive into what's the mathematics, what's the strategy. And just because 50 % of the other people in the world trying to build a quantum computer are doing something doesn't mean that it's right to me, I just don't follow that. Or I don't follow what's doing new, I'm following what I think in the long term will be right. Sometimes, most of the time this leads me in the right direction, sometimes not, and then you have to be aware of that, maybe your principles are a little bit wrong and change over time.

Himanshu Gupta (06:49.50) So thinking from first principles is a habit that from your childhood that you still carry. Are there other habits from your childhood that you still carry even now?

John Martinis (06:57.68) Yeah, the problem is it's so ingrained in yourself that you don't necessarily think about that. But I would say, you know, just perseverance. And if you have a problem, you want to work hard to try to solve it. I'm also going to say that, you know, my dad was always building things in the garage at various projects. We had a house out in the desert. He was doing things. I usually help with that. And then later on, I had my own projects, you know, being kind of trivial or in point in high school, I had electronics project and I started building things from kits and also reading textbooks of how electronics work, which is great. I can then use my math and use some principles to do that. And so building things is a really big part. And I would say having the confidence to try to build something, even it was wrong, even if it didn't make sense, just the confidence to try things.

Himanshu Gupta (07:49.76) That's an interesting point that you raised, John. The confidence to try something new doesn't matter if you fail or succeed.

John Martinis (07:55.04) Yeah, so I have a nice story on that. When I was teaching electronics at the UCSB, it's a third year course. I was very happy. It's very hard because a lot of students haven't built things and it's like building little kits. And, you know, there's a million ways to wire up something wrong and one way to wire it up right. It's as simple as saying, people have to learn that. But I was very proud when one young woman came to me one day and she said she had opened up her computer to try to change the battery and whatever. And it was the confidence of that class of playing with electronics that gave her, you know, the desire to open it up and see if she could fix it. Because she then knew just enough to do that. And what happens, of course, is once you start having the confidence to build things, then you'll try it and yeah, most of the time it's okay. And even if you break things, you're still learning something, but it's getting over that barrier that you just don't understand anything. You say, don't want to touch it. And I was really proud that that class taught that skill too. That's very important.

Himanshu Gupta (08:56.96) That's a great picturization of the problem here, right? A kid or some, even a youngster looking at a new field altogether, they feel like I don't want to touch it. Like let's say quantum or biology. And then from there, trying going to experimentation with perseverance, right? So I want to talk about the barriers there. I want to talk about those barriers. Like a lot of kids are curious. And you said like kids are born curious. What helps them get over that barrier so that they persevere in experimenting more and more?

John Martinis (09:27.40) Well, first of all, you're going to break things, okay? you know, parents, of course, could get mad that you break something. And that's kind of bad. I understand. But there's lots of stories of kids who took apart things and put it back together. you know, you try to tell your kids to start with something simple and move on to more complex. But it's okay to break things. Okay. And you know, that's part of, you know, what, what you have to do to learn, to learn how things work. And that's always a chance. Now, again, you need some guidance there. One of the interesting things, just going back to my research is there are many times, especially when we're fabricating devices, we fabricated in a new way, which is risky, you know, it's kind of this kind of thing at a very high level. And then it works out that it just doesn't work at all. And, you know, it's like, well, it should have worked and we built it a new way, but it's all broken. And then I always try to remind the students that, hey, you know, we really thought this would work. The fact that it doesn't work teaches a lot, right? You know, and that's as important as anything if it's successful because we get that feedback.

Himanshu Gupta (11:04.81) Clearly we don't want any kid watching this show to start breaking things at their homes right away. There's a method to madness, right? That's what you're talking about.

John Martinis (11:10.19) Yeah, and you should also buy a lot of EK furniture where you put it together, know, with the pictures and the like, you know, things like that. you know, I just talked to someone last week and they said they have a they have a mail order organization where they send kids a science kit every month. OK, and just get people used to doing new things, you know, different kind of science. So, you know, there's a lot of ways to encourage this.

Himanshu Gupta (11:36.75) So was there an instance in your life when you wanted to learn something new and found it extremely hard?

John Martinis (11:46.70) you know, this happens all the time in science and, what you have to do is, you know, talk to the right people, collaborate in the right way. And, you know, then understand that it's going to take a long time. I'll give you another example. I, I started rock climbing in graduate school. And you see people doing these incredible hard climbs and moves, and you realize you have no chance to do that. And then you just start training and you start with very simple things. just do easy things. You talk to friends, get people to go climbing with, have them teach you how to do that. And it's just like anything, if you want to get good at something, you just have to spend a lot of time. And I think, for example, this happens, let's say children in high school, they can join the marching band, they can do a music group or whatever. know, initially they may not be very good, but if you practice it and you're around other people and it's a good teaching environment, good learning environment, then you can do it. So there's this concept that you can get really good at something if you just put in time and persistence and do it. You may not be an Olympic athlete, most of us won't be, and you may not be the star of the football team, but that you gain a lot of knowledge and technique. And this is very important, for example, in high school, because once you realize you can get good at something by being persistent and doing something for a long time, that kind of applies to many other skills that you want to do in your life. I mean, our son just picked up our guitar and started playing it on his own. And, you know, he got fairly good at it. And it is just because he was interested in work hard. This is such an important knowledge that, you know, that by persistence, you can get good at it.

Himanshu Gupta (13:36.73) And having that realization that you can become good at something if you persevere and continue to practice and give it more time and learn from people. But in this current age, John, where the attention span is so limited, how do you give youngsters, or even college graduates, confidence that you put in that time and over a period of time, you become really good at it? How do we break that barrier?

John Martinis (14:00.46) Yeah, that's a really hard thing because people want to look at their phones and do things. I'm just going to say that generally, know, children, young adults have some hobbies and interests and, you know, look at that and, you know, try to encourage that as much as you can and, you know, understand that, you know, students getting good at something, no matter what it is, or most things, okay, but whatever that is, that's a valuable skill. In fact, that's some of the most important skills. And your children usually have some interests. Now, hopefully their interest is not social media influencers. Okay, some will do that. But I can imagine maybe that's a skill they can learn too. And then encourage them to think about what does it take to get good at something by practicing it over and over and help them, you know, help them practice it, maybe get a coach or, you know, help them by searching online to learn about how to do it.

John Martinis (15:18.77) Yeah, the interesting thing I find is if you're the best at it in your immediate circle, you know, that makes you feel good. But if you're just good at something that, you know, that has a big value, because frankly, when you grow up, the chances of you to be the best in your field is very small. Now, you're talking to someone who was awarded the Nobel Prize. But believe me, that's not why I got into physics and that's not why I loved it. I just thought it was great. It happened to you. And if it so happens that you become best, yeah, and I understand that's motivation, but you know, don't feel disappointed in yourself that you're not the best. Someone's better than you. Someone's better than you, you learn from them. Okay? So there's also a kind of mentality there that just gathering the skill is something worthwhile for you.

Himanshu Gupta (16:04.02) So in your case, clearly Nobel Prize was a byproduct, not the not star you were chasing. You just enjoyed physics.

John Martinis (16:08.72) Yeah, doing science, discovering things when you do an experiment, writing the paper, giving talks, you know, that's what's really exciting about doing science is sharing this new knowledge. You're explorer, right?

Himanshu Gupta (16:22.66) Yeah. But can I ask you, it might be a controversial thing as well, John, you talked about very interesting in science, being very good at it, but is being the best at it, right? What I see in the current field is even with PhDs or even with some of the professors, I I graduated from Stanford, is this race towards being the first at it. Race towards being the first at discovering something. Like how do you contextualize that?

John Martinis (16:45.86) Oh yeah, yes, that's definitely science. Science awards you for doing that. And there is a race. And I'm going to say that that motivates me some, but I'm going to say I want to do something well and I want to understand it. You know that. And if it's a science that requires a race, then I'm kind of a little bit not as interested in it. Although right now there's a race to build a quantum computer. So, okay, I'm part of that. wanna do it well, but I wanna do it well. Now what's very interesting, for example, is from my thesis experiment, which was the Nobel Prize experiment, but it was awarded for, there had actually been some prior papers that kind of had claimed to see it. But the problem was the papers were a little bit murky. It wasn't completely understood. they had seen what they seen. It's consistent with it, but doesn't prove it. And then what we did is try to do the experiment really cleanly so that there was very, very clear proof that we were seeing quantum tunneling and not some noise issue and the like. And also we understood the experimental physics better by understanding the microwave environment was really important. And that's why we were able to do a clean experiment, but I would say more importantly, by setting the physics knowledge and the concepts really firmly, it was clear what to do next for all the experiments and eventually led to superconducting qubits, which is a big field and why this is important thing. But I would say we put down the foundation of what the experimental physics was. And that's what was really important. So I'm gonna say in my career, I've always tried to write papers that are very clear, make sure you understand what was going on very physically. that's what was really important for me. And it turned out in this example, that was key for the Nobel Prize, for example. And that's kind of a artistic creativity kind of thing, much more than being first.

Himanshu Gupta (18:50.79) Artistic creativity. like the articulation of that term, John.

John Martinis (18:53.53) Yeah, I like that you can do that in science. It's a different way than an artist will, but you try to paint your science experiment in a creative way, a clear way. And that's something, for example, I learned very well from, you know, John Clark, my advisor, and Michelle Devereux. They really taught me well on.

Himanshu Gupta (19:11.79) Give us more examples of artistic creativity within science.

John Martinis (19:16.11) Oh gosh. There was a paper I wrote around, I think 2009 or so, on the surface code theory. So surface code is how to take physical quantum bits and do some error correction on it so that you can make error-free bits. And up until then, was many papers written on this. But in my view, it was like reading string theory. It's completely not comprehensible because they use very high level concepts and whatever. So we worked with the theorists to write a paper that was very clear that a graduate student could understand. OK, it's complicated enough that that's all we could go to. And, you know, for example, one of the key ideas of this is predicting what an error you have. qubits with physical errors, and then what would be the error corrected logical qubit error rate. And up until then, they did computer simulations, blah, blah, blah. Well, what I did is I had a very simple model where you use high school statistics. Okay, it's I think that everyone knew, and you built a little model and you use high school statistics, and then you could more or less reproduce the, you know, what you did with computer code. And, what I liked about it is at this point, you understand, you know, why the results were the way they were, because you see the formula, every understands high school statistics. So for me, this was a, a creative model that no one ever found before as simplification. And the only reason anyone believes it is because they have the exact models, but it's a way to explain something in a simple enough way that everyone could get the concept.

Himanshu Gupta (20:54.92) So the way you are defining creativity, is, and this might be my interpretation for it, is defining complicated concepts in simple ways that everyone can understand. That's one way of.

John Martinis (21:03.80) And I, you know, as a university student was exposed to this with the Feynman Lecture Series on physics, which is very well known for in the physics community that he taught a class and explained the concepts in really a different way. And a lot of times he could explain very complex concepts, advanced concepts, but by simplifying the model in some way, you could really understand what's going on.

Himanshu Gupta (21:30.68) Why does it matter, John? Especially for scientists to write papers in ways that everyone can understand. From what I am aware of, they want to write papers in a way that the committee can understand.

John Martinis (21:42.30) Well, okay, so what happens, this is kind of inside science. What happens is if you wanna get your paper through the review process, especially theoretical papers, you need to make it in some very fancy, know, mathematical way so that people think you're smart and people are enough confused by it that, okay, good. clearly something new and important. If you make it clear so that a high school or undergraduate physics student can understand, even though it's new, then they're gonna say, well, there's nothing really new here. So this is what happens with physics papers in order to get it accepted. Now, in the case that I was talking about with the surface code theory, The papers were all out there and this was kind of a review article specifically saying we want to explain this in a simple way that everyone can understand. Okay. And in that case it was fine. Oh, and by the way, that paper was accepted in, I don't know, one day by the reviewers and it has a huge number of citations. It was in the top 20 citations of physical review A for the past 50 years. So people like these papers and they'll read it, but you have to choose the right thing. So I feel sorry for theorists because they have this higher bar to cross. And frankly, if you make it very confusing, you seem very smart. Now, experimentalists, we always review theory. And we want to do that in a clear way. So it's OK for us to simplify. In some sense, that's our job.

Himanshu Gupta (23:31.17) I see. So the difference between theoretical physics and experimental physics. And you said you felt bad for a theoretical physicist, right? So do you think the way the scientific community evaluates scientific progress or innovation, does that need to change?

John Martinis (23:47.36) You know, I'm OK with the way it is because I'm complaining here, but this is just human nature. OK. And I just understand that's the way it is. I think the scientific review process is pretty good. I've always made my papers better. And, you know, when we have young people complaining about, oh, what the referee said, it's like, look, they're making our paper better. It's OK. But I would say if you do, there's ample evidence if you're doing something really fundamentally new, it's kind of harder to get it published because people don't understand. And for example, that's what we're trying to do with our company right now. And, you know, there's there's some pushback from it. And, you know, I just tell everyone, look, the fact that there's pushback means that we're trying to do something transformative. And it's just an indication that we're on the right path. Now, of course, you always have to think about what they're saying carefully.

Himanshu Gupta (25:01.19) So, but going back to our discussion at Davos, John about getting young minds interested in theoretical physics or physics in general, right? If some of the most transformative work happening in physics cannot be understand by undergraduates or even high schoolers, how do we get them interested in the next best thing in physics?

John Martinis (25:21.42) You know, this is, know, you're young and you're thinking grand thoughts. And it's this whole idea that you have to study something for a long time to get good at it. OK. And, you know, if you want to be an Olympic athlete, you have to train for years. You know, it's the same thing for a physicist. And there's a lot of things you learn. But what I would say is, you know, as you're learning the physics along the way, you know, do you enjoy it? Do you like the challenge of solving problems? I mean, that's a big deal in physics. have problem sets. And if you like solving the problems and getting better at solving problems, because that's kind of what they're teaching you the physics, but they're also teaching you new techniques to solve problems. And if you enjoy that, you know, that's kind of a good indication of what physics is like. And then eventually you see all these cool ideas that people came up with that when you get trained far enough, you might be able to discover some of those rules too. mean, that's has been a big thrill in my journey. is being able to contribute to the knowledge of physics.

Himanshu Gupta (27:36.07) So enjoying the art of problem solving is important. And by the way, we did have an Olympian athlete on the show as well. And we talked to her. And at the age of nine, she said, I want to compete in Olympics. And she's been training from the age of five to the age of 16 when she represented the United States into Olympics. So well said, John, about.

John Martinis (27:57.00) What I would say is that's a great goal. The problem is, most students, they just don't have the genes or coaches or talent or anything to do that. But it's the journey along the way of learning to be mastery. Whatever level you get to is great. The nice thing about physics or academic is you train yourself as much as can. But let's say, you you can't be a professor for some reason. That's okay. There's a lot of nice jobs out there. And so, you know, just getting good at something and, you know, figuring out how to use that is a reward in itself. That's very important.

Himanshu Gupta (28:31.72) So I want to go back John to your art of principles of explaining complicated things very simply. How would you explain your Nobel Prize winning work to a 10 year old?

John Martinis (28:42.76) Okay, that was, I thought about this a little bit. First of all, how do you explain quantum mechanics to a 10-year-old? Okay, that's what you have to start with. And, you know, one way to explain that is if they know, you know, what, how atoms are made with a proton and electron, okay, and that these attract each other. So why do atoms have size? so, you know, I would see if the 10 year old or whomever I'm talking to you can understand that that's kind of mysterious, right? Why are these, they look like point particles and yet they have size. And that's because of quantum mechanics says the electron is, I would say, I say it a little bit fuzzy. Okay, it's not a point that has some extent to it. And quantum mechanics is fundamental to the universe or else we wouldn't have matter as we see it. quantum mechanics describes how these atoms work. I could also then explain, have they ever seen a neon light? Okay, and it has a very distinct color to it. However, if you look at a regular light bulb, especially the old ones with the little filament in it, and you put current through, that gives off white light, gives all colors. So why do atoms give off a distinctive color? And that's because of quantum mechanics, the electrons, when you put current in both systems, you excite it. But with the gaseous atoms, there's only certain oscillation frequencies that the current can excite and then there's only certain colors that come out of it.

John Martinis (30:33.69) And then I would say, well, you know, I've given you examples of how atoms work. It's like kind of the physics of the small. And then what I would say what was important about my experiment is that quantum mechanics just not was for small things, but also works for electrical circuits that, you know, are about this big. And you have currents and voltages, you know, in those wires that also obey quantum mechanics. And then, you know, that's an interesting thing is that quantum mechanics has more applicability than small things. And then in the end, you know, we're trying to build this thing called a quantum computer out of it, which is what what's so interesting.

Himanshu Gupta (40:07.47) But would you recommend using AI to get answers?

John Martinis (40:12.44) Well, you have to be careful because AI could be wrong. But you know, I use AI all the time. I'm building something, and something surprising comes up. So I ask, you know, AI, what is the physics of this? what happens with this versus this? And then it quickly gives me answer to something, some subtle piece of physics that I never studied. And, I generally use two AIs to see if the results are consistent with each other. By the way, Gemini works pretty well and, you know, I use Google products, but I find Grok to be a little bit more scientific and gives me a lot of times more formulas. Okay. That, that, you know, makes sense to me, but they're different. You know, I use both.

John Martinis (40:55.69) And I'm sure there's other ones that are really good too for physics. yeah, you always have to be careful about that. But yeah, that's what I would do, I think. And also, you know, work with your child to search together. and then try to figure out, is this correct? Is it telling me the right thing? It's an exploration you can do together.

Himanshu Gupta (41:25.77) Yeah, not taking the answers from AI at the face value, but also being careful about what you ask as well.

John Martinis (41:30.30) yeah, and AI is much better than social media. so the idea that everything you read isn't necessarily the full truth. Like the big problem with social media is you tend to get only the pros. And what you want to do is get the pros and cons of any issue. And you know, I would say, yeah, you just have to be careful. But that's a general life skill that not everything people tell you and not everything you learn is going to be true and you have to be and that's again that's easier for me because I was always brought up to question everything.

Himanshu Gupta (44:19.81) I want to segue into your work at Google and you brought this up twice in the interview. And if I'm not wrong, this was the quantum supremacy experiment that you ran there. Explain to our audience, how do you define quantum supremacy?

John Martinis (44:35.05) Well, what happens, we are trying to build a quantum computer in order to run certain problems, not every problem, but certain problems much faster than you can compute on a classical computer. And the power of a quantum computer, it's kind of like a parallel processor. So for example, you can put a state into a zero plus one state. Classically, it's either zero or one. quantum mechanically can be a zero plus one. And that means you can, that one zero plus one, you can kind of get the answer of both for zero and the both of one running at once. Now, know, factor two, that's great. But when you add more qubits, like two qubits, you're actually running four states, zero, zero, zero, one, one, zero, one, one, three qubits, eight states, four, 16, five goes to 32. And by the time you get to 53 qubits, the amount of parallelization you're doing to the 53, 53 is about 10 to the 16. So it's quite long. And for that particular experiment, we had an algorithm so that you could do that parallel computation. And that the kind of computation we did, we could do in, you know, minutes or minutes hours that would take a supercomputer really, really long time to do, know, days, weeks, forever, you know, whatever. So you were able to show that that computation could be done faster.

Himanshu Gupta (49:04.90) Hmm. But why does it matter to the world?

John Martinis (49:08.47) It matters because it means in principle you can build a quantum computer that the physics will allow you to do that. Now there's good reasons to believe that's the case. You know, when you look at basic physics, that's all theoretical. I mean, you want to show that. And then, you know, I think we showed it in enough detail that, you know, people could see what the data was looking like. And it meant we had to make it better, but that was more, you know, A lot of good incremental progress needs to be made.

Himanshu Gupta (49:56.58) So in your current startup, from Google to now, what problem that you are trying to solve first? Or you are not even thinking about a problem yet?

John Martinis (50:03.84) Okay, so after quantum supremacy, I knew that we had to make the qubits better in a variety of ways. And you don't yet do that. The Google management changed at the time that they wanted to take an approach that was different than I had thought. And I stuck around for a while and it was just very clear that they wanted to do their own thing. What are we trying to do different at Colab? First of all, we're really working hard to manufacture the qubits better. Right now they're doing some process step called liftoff, which is like a stencil. You put on what's photoresist, expose it, and certain windows are then you then evaporate aluminum. And then you stick it in acetone and all the aluminum where it was, if the stencil was underneath it gets liftoff. It's very dirty. No one makes chips this way. Okay, this is just not the way to do it. So we have to make it more reliable. The other thing we have to do is we have to build the control system that controls the qubits. This is very different than standard CMOS where you have a classical control system. It's very complicated. It's large. And what we're doing is working on building that control system better. So the basic concept is we're thinking about manufacturing. And also that we're thinking not as physicists who are trying to build a quantum computer, but we're really working with the best people in the semiconductor industry who knows how to do this just with general technology. And then we're the bridge in order to say, Well, you have to take semiconductor processes and modify them in order to make better quantum bits and better control systems. So we're kind of the interface between the knowledge of the semiconductor industry and what you need to do for quantum. So in the end, it's all about manufacturing. We have to get much, much better at manufacturing qubits. And again, this is not typically how physicists think who are running the programs, especially most of the programs are run by theoretical physicists. And even the experimental physicists, they're thinking like a physicist. We have to think like an engineer, but we also have to think like a businessman who's trying to figure out how to manufacture things cheaply and reliably, like the semiconductor industry.

Himanshu Gupta (52:54.19) So you talked about the business aspect of building the supercomputer as well. So let's say in a hypothetical scenario, what will be the range of the cost we are looking at building the first quantum supercomputer, like fully operational quantum supercomputer producing real-world applications?

John Martinis (53:08.41) So the problem right now is if you just take the technology right now, what people are doing, I call it very artisanal, okay, because you build all the separate components and put it together. That machine is in the tens of billions of dollars. It's just way too expensive. Okay. And then what we're trying to do by using the semiconductor manufacturing, is make this of the cost that's roughly what supercomputers these days cost. Let's just say tens to hundreds of millions of dollars. Big data centers are more expensive. And you can only spend billions of dollars on data centers because you know what the use case is and people are making money. But generally supercomputers, let's say for research for the US government, tends to a hundreds of millions of dollars. you know, that's again, we're building a supercomputer.

Himanshu Gupta (54:30.45) Absolutely. You made an interesting point, John, about applying quantum computer to solving problems more efficiently. think of this as a parallel processor. Are there problems that a quantum supercomputer would be able to solve that are now solvable by a classical computer? We talked about problems that are currently being solved by classical computer, that a quantum computer can solve a lot better and faster. What about problems that have not been solved before so far by a classical computer. Are there some problems like that?

John Martinis (54:56.82) Well, yeah, and I'm going to say that the application that I like to talk about, and partly because I'm a physicist, is solving quantum mechanics problems from material chemistry, maybe drug discovery, things like that. Where people have solutions right now using a classical computer, it's just that they're really slow or if you get big enough, they can't do it anymore because you need too much memory or it takes too long. So there are, oh, the basic idea is you map the quantum physics of a molecule, which is of course why it's so hard to solve, to a quantum computer. And the mapping for both quantum mechanics can be done. It's not trivial, but people have understood this for quite a few years now, and there's a natural way to do that. There's also some ideas on people are using now AI and machine learning to solve these kinds of chemistry problems and that the quantum computer can help provide data, for example, to train those models in order to solve these problems well, and especially good data where it's a quantum mechanical solution. So to me, that's the really interesting things, because if we can bake better materials, then that's really important. And the one example, maybe this is a trivial example, but right now, We use rare earths to build magnets to make electric vehicles possible. But the problem with rare earths is there's a supply chain issue and they're rare enough that it's expensive and difficult, et cetera, et cetera. Well, if you could design the materials that use not so rare earths, that's maybe more ecological to mine and process and build, then that could have a huge impact. in the world. know, materials and materials discovery is a really important part of technology, especially when we want to have technology available to everyone.

Himanshu Gupta (58:19.40) How is a young kid trying to learn from AI chatbots? What should be their mode of learning? How should they evaluate what's the right way to learn versus what's not the right way to learn from an AI chatbot?

John Martinis (58:31.95) Okay, let me, I'm going to give an example of physics. That's people really have to understand a physicist, your value, unique value is able to solve problems that have never been solved before. And it could be that you have a different way to look at it. and you use some unusual mathematics or something, but If you already know how to solve it, then the engineers are really good at that, right? And they have the tools and whatever. But it's the ability to do something new and to take an abstract question and first of all, write down what the model is. Okay. It could be a simple model and then solve that. That's what we do every week when you study physics, when you're given your problem sets. There's a concept that we want you to learn about, we give you a problem set, and those problems are making you kind of invent a solution in your own mind, and you can figure out the problem if you understand the ideas. And besides understanding the ideas, you're also understanding how to solve problems better. The thing I want young people to understand is most of the problems right now that they're assigned to you, you can look it up. And if you look it up and say, yeah, now I see it and write it down, you'll turn in your problem set, but you are not learning the skill solving problems. And that's why you want to be a physicist. And that's why someone will hire you as a physicist in the long run. know, students, have to have some discipline to try to figure out how to solve the problem on your own, because that's the skill. In fact, that's the valuable skill you're learning. Not turning in the problem sets or getting an A. I mean, you're trying to figure out how to solve problems on your own. And in the end, you're going to get on your final exam. You're not going to be able to look up AI. So you have no idea to do that. So what I would say for students is resist the temptation to look it up. OK?

John Martinis (1:00:45.83) If you need to work on some of the harder problems in the study group where no one's looking it up, that's okay too. Not as good as figuring out yourself, but generally it's a social, everything's social so you can learn something. But I would say, know, as soon as you look it up, you can turn in your problem set, but you kind of have failed in your ability to like really understand how to learn problems. I understand, you know, sometimes it happens. But you have to realize you're not, it's the failing for many times and then you finally get it is where you really learn the physics. Okay, this is the end that, know, it's okay to fail for a while. If you don't know the answer right away, it's okay.

Himanshu Gupta (1:01:40.46) Yeah. learning to solve problems on your own first, Versus relying on an AI chatbot. Let's say I've tried solving a problem three times, four times, and not being able to figure it out. At what point would you recommend that student go to his or her professor? Or should they come to the chatbot?

John Martinis (1:01:55.53) Yeah, the professor is fine. I would say, you know, go to your fellow students, then go to your professor, because the professor doesn't want to feed you the answer. Right. You know, he wants that he'll have you step along. And, you know, if you can't get it, then maybe he'll feed it to you. But yes, yes, I include the professor and or the, you know, TA or whatever is there. They're gonna feed it to you because the other thing is, what's great about science is the moment of discovery. I finally understand. So we wanna feed it to you so that you can have the moment of discovery because I don't know what happens. Maybe there's like a chemical surge in your brain, dopamine surge where you get really happy that you figured it out and then you remember it, okay?

Himanshu Gupta (1:02:57.66) The hurra moment you're talking about. So do you think that AI chatbots, should not be allowed for students until a particular age?

John Martinis (1:03:07.86) Well, I'm a little bit of favor of being very careful with students with social media and chat bots and whatever and there's well-documented cases and that you know, they should be outside playing and interacting with their friends and etc. I wish there was a chat box for physics that could tell you those incremental depths so that you don't get the answer, you know right away. That would be a very nice chatbot, a physics teaching chatbot to lead you up to the solution. I think that would be very hard. And the problem is there's a lot of different misconceptions people have about physics that you have to kind of break in order for people to understand it.

Himanshu Gupta (1:04:33.08) Eventually, we will run out of true physics minds in the next two, three, four decades.

John Martinis (1:04:41.99) yeah, yeah. That's what I'm afraid if people are just looking at the solution and people will be have much less trained and less tools to do that. you know, at my age, it's great because I got trained in this way. And the other interesting thing is I was trained by physicists who were like in the Manhattan Project, who were really very good experimentalists there. So I got trained by very special generation where people were discovering all this physics, okay? So I have the best of both worlds now, but yeah, I'm afraid if people do that, but there'll always be people who learned in a different way and then they will have, you know, some advantage to it.

Himanshu Gupta (1:05:37.51) Yeah, you know, my producer is signaling that we are approaching the end of the show, John. But before we let you go, John, two quickfire questions for you. One, looking back at your life from a child living with your parents to a PhD and now Nobel laureate and a professor, what's one regret that you have that you should have learned a lot better or you should have done or a habit that you should have built earlier?

John Martinis (1:06:00.98) what I would say is, you know, I'm very singular minded and focused on physics and, you know, which, you know, is very good. I think if I would have studied a little bit more on management and, let's say psychology. and what I've learned over the years is that, you know, if you want to be even a scientific leader. You need to have a certain management style of more certainty and let's say calmness. Okay. Now, you know, again, part of that excitement or whatever that I had is I think part of the reason I'm good at the physicist, but I think understanding more of the soft people skills would have helped me allow out a lot in my career. But, you know, this also points to the fact, and I think to tell young people is in the beginning, if you focus on science or physics or whatever you're doing is great. But as you move into your career, learning these other soft skills, learning many skills is very important. And for example, for the last few years, I've been learning a lot on the tools of persuasion. Okay. And you know, we have to sell a company, right? Our startup company. So I've been learning how to do that.

Himanshu Gupta (1:07:53.86) And from what I'm getting, John is true excellence in science is not just about achieving something. It's about that ability to continue to learn new things.

John Martinis (1:08:02.49) Yeah, and let me tell you, it's great to be a professional physicist because they pay me to learn my whole career. That's your job is to learn and to get better. It's part of your job. A lot of times you have to do this a little bit on your own time or whatever, but you know, this is what's great about the job. You know, how many other jobs are like that?

Himanshu Gupta (1:08:25.69) Yeah. And the last question, John, is if there are three undergraduates in physics or even graduate in physics students who approach you, John, John, I want to learn from you and I want to become a Nobel laureate like you. How would you, who would you decide to mentor and what would you look for in those students that they have the ability to become, to, be a Nobel prize or not.

John Martinis (1:08:42.75) Well, first of all, I would say, and it just comes up in an indirect question. You do not want to have your goal in life to be a Nobel Prize recipient. It's a great honor. I love talking to students and you about it. It's fantastic, it changes your life. the chance of this happening is just really hard. There's a lot of smart people out there. And partly you have to feel be fortunate to be at the right place and right time to do something. Although in the other hand, if you want to do something, let's say transformational to your field, which could lead to a Nobel Prize, then you have to think differently. You have to notice something differently. And for example, this whole idea I said about being good at manufacturing. This is a transformational idea that very few other people appreciate at this point. And so I'm being very direct about it. This is something that sounds a little bit crazy. I mean, to you and I, maybe it's okay, but the people in the field, I've talked to them. People are not interested in what we're doing. They think they're totally fine. And maybe they're right. But it's more of an attitude that you. You think of something transformational, doesn't make sense, you approach it, there's a risk involved. And yeah, that's the way I would talk to people about that. And you just have to keep your eyes and ears open to look for new ideas that maybe not everyone is excited about, but could do something. And then you have to have good ability to separate out whether it's a good idea or not and try it.

Himanshu Gupta (1:10:18.27) So aiming for novelty is more important than aiming for nobel. And I think

John Martinis (1:10:22.07) I would say aiming for a transformational in your field. Okay? That's the way I think about it.

Himanshu Gupta (1:10:32.02) That's a great way to cap our conversation as well, John. Cannot thank you enough for imparting your wisdom to our audience. And I'm sure you will inspire many young kids to get into physics.